Blog AI & ML

Moving Toward More Efficient Networks with Machine Learning

Carolina Bessega Innovation Lead, Office of the CTO Published 8 Dec 2022

Infographic Series - Efficient Networks and Machine LearningIn the next few years, all industries will have fully adopted a combination of various data/network-hungry technologies, such as augmented reality (AR), virtual reality (VR), and artificial intelligence (AI). Some examples that we start to see today are predictive maintenance, precision farming, intelligent energy grid automation, robot assistance, and immersive educational experiences.

According to Statista, the number of IoT devices forecasted for 2030 is more than 29 billion. Surprisingly (or not), this is more than 3.5 times the current earth population. These devices will contribute to massive global data creation and consumption, projected to be more than 180 zettabytes by 2025. In parallel, adopting new technologies such as VR with 360° immersive experiences imposes ultra-low latency and ultra-high throughput requirements on mobile networks. If we want to achieve a resolution similar to a 4K television, VR devices will require a bandwidth of 400 Mbps with latency below 20 ms.

Emerging network technologies like 5G and Wi-Fi 6E and their successors (6G and Wi-Fi 7) hold promise for solving significant problems related to network capacity, latency, and reliability. These wireless technologies are key enablers in scaling the previously mentioned use cases.

Network function virtualization and software-defined networking are becoming the standard to add flexibility and optimize network operations. Still, with the increasing technological demand, automation and intelligent network operations leveraging machine learning (ML) will be critical in the years to come to reach the desired quality of service (QoS) and achieve the desired customer quality of experience (QoE). Additionally, ML will assist in reducing the operational costs of telecommunications.

A few examples of machine learning (ML) techniques in use or research phases for communication networks include:

  • Reinforcement learning for optimal resource allocation
  • Session scheduling or routing in dynamic networks
  • Unsupervised anomaly detection for QoS and QoE
  • Convolutional Neural Networks (CNN) for indoor user localization
  • K-Nearest Neighbor (KNN) and support vector machine for interference prediction or antenna selection

We have chosen the OSI Model to illustrate how important machine learning is becoming interwoven at every layer of network communications. The examples visualized in the infographic do not constitute the totality of the possibilities.  We want to open a dialogue to ask, “what would be the most impactful use of machine learning in communication networks in the next 5 years?” I hope to engage with you as we try to answer that question together. Please check back soon for more blogs and content about the possibilities of machine learning within communication networks.

You can grab the full PDF version of this ML and OSI model infographic via the link below. So, please take a moment to download the PDF, print it out in full color, and display it in your office.

Download today!

Get the latest stories sent straight to your inbox!

Related Stories